When you think „mountain runner”, who do you see? A gaunt ultra racer at three in the morning, kilometer one hundred and twenty, the lead pack, hallucinations? I used to think so too. Then I sat down over the largest publicly available dataset of mountain-running results — 2.96 million results from 15.5 thousand races worldwide, gathered for an academic paper on trail-running performance prediction¹ — and worked out what it really looks like.
A few numbers from this analysis changed how I think about who I'm building UltraFlow for. They might change yours too — about where you sit on that map.
Myth one: „real mountain running starts at 100k”
It doesn't. 90% of all mountain-race starts fall between 15 and 110 kilometers. Races of 110 km and longer — the ones films get made about — are 7.5% of starts. A niche within a niche.
The single biggest band is 45–70 km — „short ultra”, a mountain marathon with a twist. And races of 15–45 km — skyrace, short trail — are together 40% of the world's starts. So the statistical mountain runner doesn't run hundreds. They run a 25k with two thousand meters of climbing — and let's be honest, that can be harder than a flat hundred.
Why does it matter? Because terrain obeys the same laws regardless of distance. A descent wrecks your legs on a 25 km skyrace just like on a 150 — you just find out faster.
Myth two: „the mountains are a men's sport”
In aggregate — 79% of starters are men. But watch what happens as the distance shortens: at races up to 15 km, women are 42% of starters. At 15–30 km — 29%. The share drops with distance, to 16% at 70–110 km.
Women's trail running is alive and well — just not where the cameras point. It lives on the short, intense, technical distance. One more reason to stop equating „mountain running” with hundreds.
Poland: the world's eighth power (seriously)
I didn't expect this one. The dataset holds 29 thousand Polish mountain runners — eighth place in the world, ahead of Switzerland and Germany. Ahead of Germany — a country four times more populous, with the Alps at its doorstep.
And two Polish quirks the data shows:
We run longer than the world. Distances of 45–110 km are 58% of Polish starts (world average: 50%). A Pole in the mountains picks Bies over a 15k more often.
We race more often. In the last full year of the data, 39.5% of active Polish runners logged at least two starts (world: 30%). The statistical Polish mountain runner does ~3 starts — more than an American (2.4) or a Brit (2.2).
And the data ends in 2018 — when Polish trail was only picking up speed. Growth within the set alone: from 154 starts a year to over 30 thousand in eight years. Two hundredfold.
How long „should” it take? The population answers
One of the more interesting things you can compute from a set like this: finish-time distributions for route classes. Not records, not the elite — the middle of the pack, ordinary people like us. Examples (median finish):
What's it for? Because the first question with any goal is „how long will it take me” — and the honest answer at the start of a build is: as long as the population on a similar route, until your own data says something more precise. That's exactly how UltraFlow works: the system starts from an honest band drawn from published research and narrows it to you as it learns your long runs. No crystal-ball prophecy down to the minute — a band with a source.
The most important number: a season, not a race
Of all the numbers, this one matters most to me: one in three mountain runners worldwide — and in Poland closer to one in two-to-three — starts more than once a year. Which means they aren't preparing for a race. They're running a season: a spring 30k as an opener, a summer 70k, an autumn main goal. Three tapers, three recoveries, transitions between blocks, windows where the body lies that it's ready.
And that's exactly the moment where „I have a watch and a plan off the internet” stops being enough — because a plan off the internet doesn't know you finished an ultra two weeks ago and your tendons are halfway through their rebuild, even though HRV is back to normal.
Caveats (because numbers without caveats are marketing)
The data covers 2010–2018 — the market has clearly grown since, so the proportions are more reliable than the absolute values. The set is built from an international race database, which favors larger, certified events. Runner uniqueness is estimated approximately. Treat it as a map of the terrain, not a notarized deed.
What I'll do with it
From day one I built UltraFlow with „for mountain runners, from skyrace to miler” in mind — and these numbers say that wasn't wishful thinking, it was a description of the market. 90% of starts within reach of one system that understands a VK and a 150 km need different plans — and picks the right one from the route profile.
Running a season, not a single race?
UltraFlow is live — it takes you from today to the start line, and knows the difference between a VK and a 150.
Join →AI reports · for fitness and performance only · not a substitute for a certified coach
And in six days I'll check my own numbers over 110 kilometers. Because the best test of a system is your own skin.
Tomasz · UESCA Certified Ultra Coach · Founder, UltraFlow™
Firefighter, Polish State Fire Service. Mountain ultra runner. SOG 160 km, Bieg Ultra Granią Tatr 71k, Mityczny Bies 135k.
¹ Data: Fogliato R., Oliveira N., Yurko R., TRAP: A Predictive Framework for the Assessment of Performance in Trail Running, Journal of Quantitative Analysis in Sports (2020); the publicly released dataset accompanying the paper (2.96M results, 15,527 races, 2010–2018). All statistics in the text are the author's own aggregates from this set.